JP4688862B2 - Providing support for single step functionality of virtual machines in virtual machine environments - Google Patents

Description

  Embodiments of the present invention generally relate to virtual machines. Embodiments of the present invention are particularly concerned with providing support for a single step function of a virtual machine in a virtual machine environment.

Single-step technology is widely used by debuggers in conventional operating system (OS) environments. In particular, single-step technology is used to advance an application one instruction at a time, thereby allowing the debugger to place the application at a specific point in the source code. Various techniques have been used to allow single step. For example, in the Intel® Pentium® 4 instruction set architecture (ISA) (referred to herein as the IA-32 ISA), the trap flag (TF) in the EFLAGS register allows the debugger to Specified to allow single-stepping. If the TF bit is set to 1, a debug exception is generated after completion of the next instruction. After the debugger gains ownership of the debug exception, the debugger sets the TF bit (eg, by creating a handler for the exception and causing the handler to be called in the event of a debug exception). However, a debug exception is generated only when the next instruction ends normally. If execution of the next instruction causes a fault (eg page fault), no debug exception is generated. Rather, the exception is vectorized, storing the value of the EFLAGS register bit and clearing the TF bit. When the handler is complete, the stored value of the TF bit is restored and the instruction is re-executed. If no fault occurs on re-execution, a debug exception is generated.
US Pat. No. 5,522,075 US Pat. No. 6,397,242 US Pat. No. 6,968,806 U.S. Pat. No. 7,111,200 US Patent Application Publication No. 2005/0081199 US Patent Application Publication No. 2005/0076186 US Patent Application Publication No. 2005/0091652 US Patent Application Publication No. 2005/0223221

  The present invention is illustrated by the use of the accompanying drawings, in which reference numerals refer to similar elements, as examples of the invention, which are not intended to limit the invention.

1 is a diagram illustrating an embodiment of a virtual machine environment in which the present invention can operate.

FIG. 6 is a flow diagram illustrating one embodiment of a process for providing support for a virtual machine single step function in a virtual machine environment.

FIG. 6 is a flow diagram illustrating one embodiment of a process for responding to a VM entry request with a single step indicator set to a single step value.

FIG. 6 is a flow diagram illustrating one embodiment of a process for processing an event having a higher priority than a single step.

FIG. 6 is a flow diagram illustrating an exemplary process using a single step of a VM to facilitate device virtualization, according to one embodiment.

  A method and apparatus for providing support for single step functionality of a virtual machine in a virtual machine environment is described. In the following description, for the purposes of illustrating the present invention, numerous specific details are set forth in order to provide an understanding of the present invention. However, it will be apparent to one skilled in the art that the present invention may be practiced without these details.

  Some portions of the detailed descriptions that follow are presented in terms of algorithms and symbolic representations of operations on data bits within a computer system register or memory. These algorithmic descriptions and representations are the methods used by those skilled in the data processing arts to most effectively convey the substance of their technology to others skilled in the art. Algorithms are also considered herein to be a self-consistent sequence of actions that lead to expected results. The operations are those requiring physical operations on physical quantities. Usually, though not necessarily, these quantities take the form of electrical or magnetic signals capable of being stored, transferred, combined, compared, and otherwise manipulated. Referencing these signals in bits, values, elements, symbols, letters, words, numbers, or any such form has often proved useful, mainly because of their widespread use. ing.

  However, it should be understood that the above and similar terms are associated with appropriate physical quantities and are merely convenient labels applied to these quantities. Unless otherwise stated, as will be apparent from the description below, the terms “processing”, “computing”, “calculation”, “determining”, or similar terms used in the description relating to the present invention are Physical in a device that manipulates data expressed as physical (electronic) quantities in system registers and memory to store, transfer, or display computer system memory or registers, or other information Refers to the operation and processing of a computer system or similar electronic computing device that translates into other data that is also expressed as a small quantity.

  In the following detailed description of the embodiments, reference is made to the accompanying drawings that show specific embodiments in which the invention can be practiced, as a means of describing the invention. In the accompanying drawings, numerals and the like describe components that are substantially similar throughout in several figures. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention. Other embodiments may be used and structural, logical, and electrical changes may be made without departing from the scope of the invention. Further, it should be understood that the various embodiments of the present invention are different but not necessarily mutually exclusive. For example, a particular function, structure, or property described in one embodiment can be included in other embodiments. The following detailed description is, therefore, not to be taken as limiting the invention, and the scope of the invention is limited only by the appended claims and the equivalents of the claims. Defined.

  The following example describes a single step of a virtual machine in connection with execution units and logic circuits, but other embodiments of the invention can be implemented by software. For example, in some embodiments, the present invention provides a machine or computer that stores instructions that may be used to program a computer (or other electronic device) for the purpose of performing the processing according to the present invention. It may be provided as a computer program product or software that includes a readable medium. In other embodiments, the methods of the present invention may be implemented by specific hardware components that include hardware controlled logic or programmed computer components and customs intended to perform the methods of the present invention. -It may be executed by a combination of hardware components.

  Thus, machine-readable media may include any mechanism for storing or transmitting information in a form readable by a machine (eg, a computer). These mechanisms are floppy disk, optical disk, compact disk read only memory (CD-ROM), magnetic optical disk, read only memory (ROM), random access memory (RAM), erasable Programmable read-only memory (EPROM), electrically erasable read-only memory (EEPROM), magnetic or optical card, flash memory, Internet transfer, electrical, optical, acoustic, or other forms of propagation signals ( Example: carrier wave, infrared signal, digital signal, etc.), but not limited to.

  Furthermore, the design may go through various stages from creation to simulation and manufacturing. Data representing a design can represent the design in many ways. Initially, the hardware may be expressed using a hardware description language or other functional description language, as is convenient in simulation. In addition, circuit level models with logic and / or transistor gates may be created at several stages of the design process. Many designs also reach a level of data that represents the physical placement of various devices in the hardware model at some stage. When conventional semiconductor manufacturing techniques are used, the data representing the hardware model may be data specifying the presence or absence of various features in different mask layers of the mask used to manufacture the integrated circuit. In all design representations, the data may be stored on any form of machine-readable media. Light waves or radio waves that are adjusted or generated to transmit these information, magnetic or optical storage such as memory, or disks may be machine-readable media. All of these media may “transmit” or “represent” design or software information. When an electrical carrier that represents or transmits a code or design is transmitted, a new copy is created in such a way that copying, buffering, or retransmission of the electrical signal is performed. Thus, a communications provider or network provider may generate a copy of an object (carrier) that embodies the techniques of the present invention.

  FIG. 1 is a diagram illustrating one embodiment of a virtual machine environment 100 in which the present invention can operate. In this embodiment, the minimal platform hardware 116 comprises a computer platform capable of running a standard operating system (OS) or a virtual machine monitor (VMM) such as the VMM 112, for example.

  The VMM 112 is generally implemented in software, but may emulate and export a minimal machine interface to the upper software. Such higher-level software may include a normal or real-time OS, may be a largely removed operating system having limited operating system functions, and may have conventional OS functions. It does not have to be. Alternatively, for example, the VMM 112 may be executed in or on another VMM. A VMM may be implemented, for example, by hardware, software, firmware, or a combination of various techniques.

  Platform hardware 116 may be a personal computer (PC), mainframe, handheld device, portable computer, setup box, or any other computer system. Platform hardware 116 includes a processor 118 and memory 120.

  The processor 118 may be any type of processor capable of executing software, such as a microprocessor, digital signal processor, microcontroller, and the like. The processor 118 may include microcode, programmable logic, or hardware controlled logic for the purpose of performing the methods of the embodiments of the present invention. Although FIG. 1 shows only one processor 118, there may be more than one processor in the system.

  Memory 120 may be a hard disk, floppy disk, random access memory (RAM), read only memory (ROM), flash memory, a combination of the above devices, or any other type of machine medium readable by processor 118. It may be. The memory 120 may store instructions and / or data that perform the methods of the embodiments of the invention.

  The VMM 112 provides one or more virtual machine (VM) abstractions to other software (ie, “guest” software). One or more VMs may provide the same or different abstractions for different guests. FIG. 1 shows two VMs 102 and 114. Guest software executed on each VM may include a guest OS, such as guest OS 104 or 106, and various guest software applications 108 and 110. Each of the guest OSs 104 and 106 has access to physical resources (eg, processor registers, memory, and I / O devices) in the VMs 102 and 114 on which the guest OS 104 or 106 is running, and other functions. Assume to execute. For example, assume that the guest OS has access to all registers, caches, structures, I / O devices, memory, etc., depending on the processor architecture and platform provided in the VM. These resources accessed by the guest software can be classified as “privileged” or “non-privileged”. For privileged resources, the VMM 112 facilitates the functions required by the guest software while maintaining full control of these privileged resources. Non-privileged resources do not need to be controlled by VMM 112 and can be accessed by guest software.

  In addition, each guest OS has exceptions (eg, page faults, general protection violations, etc.), interrupts (eg, hardware interrupts, software interrupts), and platform events (eg, initialization (INIT) and system management interrupts (SMI)). Assume that various events such as)) are processed. Some of these fault events are “privileged” because they need to be handled by the VMM 112 for the purpose of ensuring the normal operation of the VMs 102 and 114 and protecting them from the guest software. is there.

  When a privileged fault event occurs or when guest software accesses a privileged resource, control can be transferred to the VMM 112. The transfer of control from the guest software to the VMM 122 is referred to as VM exit in this specification. After facilitating resource access or processing the event appropriately, the VMM 112 may return control to the guest software. The transfer of control from the VMM 112 to the guest software is referred to as VM entry in this specification.

  In one embodiment, processor 118 controls the operation of VMs 102 and 114 according to data stored in virtual machine control structure (VMCS) 124. The VMCS 124 may include the status of the guest software, the status of the VMM 112, execution control information indicating how the VMM 112 wants to control the operation of the guest software, information for controlling the transition between the VMM 112 and the VM, and the like. The processor 118 reads information from the VMCS 124 for the purpose of determining the execution environment of the VM and suppressing its operation. In one embodiment, the VMCS is stored in the memory 120. In some embodiments, multiple VMCS structures are used to support multiple VMs.

  In one embodiment, the processor 118 receives a request for the VMM 112 to transfer control to the VM 102 or 114 (request for VM entry) and determines whether the VMM 122 has requested a single step for the associated VM. It has logic 122. The single step function may have various uses in the virtual machine environment 100. For example, the VMM 112 may require single-step functionality for the purpose of debugging system (ie, privileged) or application code running on the VM 102 or 114. In other examples, the VMM 112 may require single-step functionality for the purpose of achieving virtualization of a particular device or platform during operation of the VM 102 or 114 (eg, the VMM 112 may read processing from the VM). Memory-mapped I / O devices that do not want to allow write processing). In yet another example, the VMM 112 may require a single-step function for the purpose of tracing the execution of the VM 102 or 114 (maintaining the course of VM operation when testing new system designs). For).

  The single step logic 122 determines whether the VMM 122 has requested a single step function based on the current value of the single step indicator. In one embodiment, the single step indicator is stored in VMCS 124. Alternatively, the single step indicator may be present in processor 118, memory 120 and processor 118 combination, or any other storage location.

  In one embodiment, the VMM 112 sets the value of the single step indicator before requesting the VM 102 or 114 to transfer control. Alternatively, each of VMs 102 and 114 is associated with a different single step indicator that is set to a predefined value or changed during operation of the VM.

  When the single step logic 122 determines that the single step indicator is set to a single step value (eg, 1) upon receipt of the VM entry request by the VMM 112, the single step logic 122 When the first VM instruction is executed and the execution of the first VM instruction is normally completed, the control is transferred to the VMM 112 (that is, the VM exit is generated). In one embodiment, the single step logic 122 notifies the VMM 112 that the VM exit was triggered by the current value of the single step indicator (eg, using a specified reason code).

  If execution of the first VM instruction fails (ie, caused a fault), the single step logic 122 examines the associated execution control information in the VMCS 124 (eg, the fault is VM Determine whether the resulting fault is related to a VM exit (by determining whether it indicates that exit is required). If the fault is associated with a VM exit, the single step logic 122 transfers control to the VMM 112 and, in one embodiment, notifies the VMM 112 that the VM exit was caused by the fault. If the fault does not require a VM exit, the single step logic 122 sends the fault to the VM. In one embodiment, sending a fault retrieves the redirect structure of the entry associated with the sent fault, and extracts from this entry the descriptor of the routine location designated to handle the fault; Using descriptors to move to the beginning of the routine. Routines designated to handle corresponding interrupts, exceptions, or other faults are referred to herein as fault handlers.

  During sending of the fault, the processor 118 may perform address conversion for converting an address from a virtual address to a physical address one or more times. For example, the interrupt table address or the associated handler address may be a virtual address. The processor may need to perform various checks during fault delivery. For example, the processor can check segmentation registers and access addresses (resulting in restriction violation faults, absent segment faults, stack faults, etc.) and permission levels that can result in protection violations (eg, general protection violations) Consistency confirmation, such as confirmation, may be performed.

  Address translation and verification during the fault vectorization process can result in various faults such as page faults, general protection faults, and the like. Some faults that are triggered during the sending of the current fault may cause a VM exit. For example, if the VMM 112 requires a VM exit in the event of a page fault, for the purpose of protecting and virtualizing physical memory, a page fault that occurs while sending the current fault to the VM will result in the VM exit being Arise.

  In one embodiment, the single step logic 122 addresses the possibility of the additional faults described above by checking whether the current fault delivery was successful. If the delivery of the current fault is successful, the single step logic 122 generates a VM exit before executing the fault handler instruction and indicates that the VM exit was generated by the current value of the single step indicator. , Notify the VMM 112. If the single step logic 122 determines that the delivery has failed, the single step logic 122 repeats the above process for the new fault.

  For certain instructions, successful completion of the instruction includes fault, exception, or interrupt vectorization. For example, in the IA-32 ISA, software interrupt instructions (ie, INTn, INTO, INT3) vectorize software interrupts by generating calls to interrupts or exception handlers. When such an interrupt occurs, the VM exit is generated by the single step mechanism before the first instruction of the interrupt or exception handler in the VM is executed (no nested exception occurs while sending the software interrupt). Also assume that a VM exit does not occur due to a fault or exception).

  It should be noted that the VM may use a debug or single step mechanism that does not rely on the single step mechanism described herein. For example, in an IA-32 ISA, the VM may set the TF bit in the EFLAGS register to 1 for the purpose of generating a single-step trap, or for the purpose of generating a debug breakpoint Debug registers may be used. If the first VM instruction causes such a breakpoint or trap, a VM exit is generated by a single step mechanism before vectorizing the breakpoint or trap. Information about wait breakpoints or traps is stored in the VMCS, in one embodiment, for use by the VMM to properly emulate the mechanism used by the VM.

  FIG. 2 is a flow diagram illustrating one embodiment of a process 200 that provides support for single step functionality of a VM in a VM environment. This processing can be hardware (eg, circuitry, dedicated logic, programmable logic, microcode, etc.), software (software running on a general purpose computer system or a dedicated machine), or processing logic having a combination of hardware and software May be executed by In one embodiment, process 200 is performed by single step logic 122 of FIG.

  Referring to FIG. 2, process 200 begins at process block 202 where processing logic receives a request from a VMM to transfer control to a VM. In one embodiment, the request to transfer control is received by a VM entry instruction executed by the VMM.

  In decision box 204, processing logic determines whether the VMM requested a single step of the VM being invoked. In one embodiment, processing logic determines whether the VMM has requested a single step of the VM by reading the current value of the single step indicator maintained by the VMM. The single step indicator may be present in the VMCS or in any other data structure accessible by the VMM and processing logic 200. In one embodiment, if the VMM wants to perform a single step of the VM, the VMM sets the single step indicator to a single step value (eg, 1) and generates a request to transfer control to this VM. In one embodiment, if the VMM sets the single step indicator to a single step value, the VMM also sets an execution control indicator so that a VM exit occurs for each fault that can occur during execution of the VM. . Alternatively, the setting of the single step indicator does not depend on some or all execution control indicators.

  In one embodiment, the single step indicator is reset to a non-single step value (eg, 0) at each VM exit. In other embodiments, if the VM single step is not required, the VMM resets the single step indicator to a non-single step value before making a request to transfer control to the VM.

  As described above, for example, when debugging system or application code running on a VM, virtualizing a device or platform while the VM is running, tracing VM execution, etc. A single step may be requested.

  If the processing logic determines that a single step of the VM is required, the processing logic executes the first instruction of the VM at processing block 208 and the execution of the first VM instruction is normal in decision box 210. Determine if finished. If execution of the first VM instruction is successful, processing logic generates a VM exit at processing block 212 and notifies the VMM that the cause of the VM exit is a single step at processing block 214. In one embodiment, processing logic notifies the VMM of the cause of the VM exit by generating a specified reason code. In one embodiment, the reason code is stored in the VMCS.

  If execution of the first VM instruction fails, processing logic determines in decision box 216 whether the resulting fault causes a VM exit. This determination is made by examining associated execution control information (eg, stored in VMCS 124). If the resulting fault generates a VM exit, processing logic generates a VM exit at processing block 222 and notifies the VMM at processing block 224 that the cause of the VM exit is a fault. If the resulting fault does not cause a VM exit, processing logic sends the fault to the VM at process block 218 and a determination box 220 determines whether the sending of the fault is successful. If the delivery of the fault is successful, processing logic generates a VM exit before executing the corresponding fault handler instruction in processing block 212, and in processing block 214 the cause of the VM exit is a single step. This is notified to the VMM.

  If the fault delivery fails, processing logic returns to decision box 216 for the purpose of processing the resulting new fault.

  If the processing logic determines at decision box 204 that a single step of the VM is not required, processing logic executes the first instruction of the VM at processing block 226 and the first VM instruction at decision box 228. Whether or not the execution of has been completed normally. If execution of the first VM instruction is successful, processing logic executes the next instruction of the VM at process block 230 and returns to decision box 228. Otherwise, if execution of the first VM instruction fails, processing logic determines in decision box 232 whether the resulting fault causes a VM exit. If the fault causes a VM exit, processing logic generates a VM exit at processing block 222 and notifies the VMM at processing block 224 that the cause of the VM exit is a fault. If the fault does not cause a VM exit, processing logic sends the fault to the VM at processing block 234 and a decision box 236 determines whether the sending of the fault is successful. If the fault delivery is successful, processing logic executes the first instruction of the fault handler and returns to decision box 228. If the fault delivery fails, processing logic returns to processing block 234 for the purpose of processing the resulting new fault.

  In some embodiments, certain events may occur during the operation of a VM invoked in response to a request for a VM entry associated with a single step indicator set to a single step value. For example, the VMM requires that a fault be sent to the VM as part of the VM entry, that the VM entry put the processor in an inactive state, that the first instruction after the VM entry causes a VM exit, etc. It is good. In the Intel Pentium 4 instruction set architecture (ISA) (referred to herein as the IA-32 ISA), examples of inactive states include Halt, Wait-for-SIPI, shutdown, and MWAIT state. During these inactive states, the processor does not complete any instructions and waits for the occurrence of one or more break events that can cause the processor to transition from the inactive state to the normal state. For example, during a Halt active state, the processor may not block hardware interrupts, system management interrupts (SMI), system initialization messages (INIT), non-maskable interrupts (NMI), debug exceptions, machine confirmation exceptions, etc. Wait for outbreak. A typical process for handling such events while supporting the VM's single-step functionality is described below in conjunction with FIGS.

  FIG. 3 is a flow diagram illustrating one embodiment of a process 300 for responding to a request for a VM entry with a single step indicator set to a single step value. This process can be hardware (eg, circuitry, dedicated logic, programmable logic, microcode, etc.), software (software running on a general purpose computer system or a dedicated machine), or a combination of hardware and software It may be performed by logic. In one embodiment, process 300 is performed by single step logic 122 of FIG.

  Referring to FIG. 3, process 300 begins at process block 302 where processing logic detects a VM entry with a single step indicator set to a single step value. Next, in decision box 304, processing logic determines whether the VMM has requested delivery to the standby fault VM as part of the VM entry (i.e., whether the vectorized fault is sent to the VM). please). If so, processing logic sends a wait fault to the VM at processing block 306. Fault delivery searches the entry's redirect structure associated with the fault being sent, extracts the corresponding fault handler location descriptor from this entry, and uses this descriptor to It may include moving to the top of the handler. While sending faults, some actions that could result in new faults (eg page faults, general protection violations, etc.) (eg address translation, segmentation register and access address checks, permission level checks, Etc.) may need to be performed. In decision box 308, processing logic determines whether the sending of the standby fault has been completed successfully. If so (ie, if a new fault has not occurred), processing logic, at processing block 310, executes VM exit with a reason code that specifies a single step indicator before executing the instruction of the corresponding handler. Is generated.

  If delivery of the wait fault fails, processing logic determines in decision box 312 whether the resulting new fault causes a VM exit. If so, processing logic generates a VM exit at process block 314 with a reason code specifying the new fault. Otherwise, if the new fault does not result in a VM exit, processing logic returns to processing block 306 for the purpose of handling the new fault.

  If the processing logic determines in decision box 304 that the VMM has not requested the sending of a standby fault as part of the VM entry, then processing logic further determines that the VM entry inactivates the processor in decision box 316. Determine whether to transition to the state. If so, processing logic transitions to an inactive active state (eg, a Halt state) at processing block 317, and transitions the processor from an inactive active state to a normal (ie active) active state at processing block 318. Wait for a break event that can be made, and in decision box 320, determine whether the break event causes a VM exit (eg, by examining relevant execution control information in VMCS 124). If this determination is positive, processing logic generates a VM exit at process block 322 with a reason code specifying a break event. In some embodiments, if the VM's operating state indicator needs to be stored as part of the VM exit (determined, for example, by examining a stored operating state indicator in execution control information in the VMCS). The indicator of the operating state (that is, the inactive active state) before the occurrence of the break event is saved. Otherwise, if the break event does not result in a VM exit, processing logic proceeds to processing block 306 and attempts to send the break event to the VM. In one embodiment, an attempt to send a break event to the VM causes the VM to transition from an inactive active state to a normal (ie active) active state, and if the VM operational state indicator is saved, the indicator is It has the effect of indicating a normal (ie active) activity state.

  If the processing logic determines in decision box 316 that the VM entry does not cause the processor to transition to an inactive active state, processing logic executes the first instruction of the VM in processing block 324 and the first in decision box 326. Whether execution of the instruction causes a VM exit. If so, processing logic generates a VM exit at process block 328 with a reason code identifying the instruction. Otherwise, processing logic determines in decision box 330 whether the first instruction was successful.

  If execution of the first instruction fails, processing logic proceeds to decision box 312 for the purpose of handling the resulting new fault. If execution of the first instruction is successful, processing logic further determines in decision box 332 whether the instruction causes the processor to transition to an inactive active state. For example, in an IA-32 ISA, an HLT instruction causes the processor to transition to a halt state, which is an inactive active state. If the executed instruction actually transitions to the inactive active state, processing logic further determines in decision box 334 whether an active indicator needs to be saved (eg, execution control in VMCS). Determine by examining the information storage activity status indicator. If the activity indicator needs to be saved, processing logic saves the activity indicator (eg, in the VMCS) at process block 336 and specifies a single step indicator at process block 338. A VM exit is generated together with the reason code to be executed. If the instruction does not cause the processor to transition to the inactive active state, or if the instruction causes the processor to transition to the inactive active state but the active indicator does not need to be saved, processing logic may Proceed directly to 338.

  In some embodiments, a single step may have a priority that is lower than the priority of some other events (eg, platform events such as INIT and SMI). FIG. 4 is a flow diagram illustrating one embodiment of a process 400 for processing events having a higher priority than a single step. This process can be hardware (eg, circuitry, dedicated logic, programmable logic, microcode, etc.), software (software running on a general purpose computer system or a dedicated machine), or a combination of hardware and software It may be performed by logic. In one embodiment, process 400 is performed by single step logic 122 of FIG.

  Referring to FIG. 4, process 400 begins at process block 402 where processing logic detects a VM entry with a single step indicator set to a single step value, and at process block 404 the first instruction of the VM is Execute.

  Next, after execution of the first instruction of the VM, processing logic detects an event (eg, NMI, SMI, etc.) having a higher priority than a single step in processing block 406, and in decision box 408, the priority Determine whether a high event is handled by the VMM or some specified software. In one embodiment, if a VMM that uses a single-step mechanism can set the execution control indicator for a high priority event to cause the VM exit to occur before the associated VM entry, the priority is high. Events are processed by the VMM. If a VMM using a single-step mechanism cannot set the corresponding execution control indicator as described above (eg, if no such execution control indicator is available), the high priority event is Processed by the specified software. For example, a VMM that uses a single step mechanism may set the corresponding execution control bit so that the NMI triggers the VM exit prior to the associated VM entry. In another example, a VMM that uses a single step mechanism may not have the capability to update the execution control indicator associated with the SMI before the associated VM entry. The SMI may then be processed by software designated to manage high priority events.

  If a high priority event is processed by a VMM using a single step mechanism, processing logic generates a VM exit at process block 410 with a reason code specifying the high priority event. Alternatively, if a high priority event is processed by software designated to manage the high priority event, processing logic sets a waiting VM exit indicator (eg, in VMCS) at processing block 412. The single step value is set, and control passes to designated software at process block 414. The designated software then processes the high priority event at block 416 to facilitate delivery of the standby VM exit to the VMM through a single step mechanism. The standby VM exit indicator may be sent to the VMM by designated software or by the processor when processing of a high priority event is complete. For example, the SMI process may be processed by management software independent of the VMM. When the SMI processing is completed, the management software may execute an RSM (return from system management mode) instruction, and when executing this instruction, the processor recognizes the presence of the standby VM exit by a single step mechanism. The standby VM exit may be sent accordingly.

  As described above, VM single stepping with VMM can include a variety of, including, for example, debugging of system or application code executed in VM, device or platform virtualization during VM operation, tracing of VM execution, etc. You may have a use. A typical process using single step for the purpose of facilitating device virtualization is described in detail below.

  FIG. 5 is a flow diagram illustrating an exemplary process 500 using a single step of a VM aimed at facilitating device virtualization, according to one embodiment. The device may be an input / output (I / O) device that uses a memory mapped programming model (eg, highly programmable interrupt controller (APIC)), and the VMM is a memory mapped I / O by VM. It may be required to allow the VM to read data from the memory mapped I / O device while preventing data writing to the device. Process 500 includes hardware (eg, circuitry, dedicated logic, programmable logic, microcode, etc.), software (software executing on a general purpose computer system or a dedicated machine), or a combination of hardware and software. It may be performed by logic.

  Referring to FIG. 5, process 500 sets the VM's page table for processing logic in the VMM to write-protect the memory mapped space of the device (eg, using the corresponding execution control information). Beginning at process block 502, a page fault is set such that a page fault that occurs during execution of the VM causes a VM exit. In one embodiment, the VM page table configuration uses a virtual translation lookaside buffer (VTLB) data structure that maintains an independent active guest page table for each VM under the control of the VMM. Enabled by.

  At processing block 504, processing logic within the processor generates a page fault VM exit when the VM accesses the memory mapped space of the device.

  At processing block 506, processing logic within the VMM causes the VM's access to the protected memory mapped space (eg, by investigating that the fault address value fits within the memory mapped I / O page). Judge that it was a thing.

  At processing block 508, processing logic within the VMM is responsible for mapping the fault address to a different physical memory page that exists in normal physical memory (ie, not memory mapped I / O space). Change the table.

  At processing block 510, processing logic within the VMM sets a single step indicator and requests a VM entry.

  At processing block 512, processing logic within the processor executes the VM's instructions for the purpose of accessing the page selected by the VMM.

  At processing block 514, processing logic within the processor generates a VM exit with a single step indicator after successful completion of the instruction.

  Next, processing logic within the VMM examines the data written to the physical memory page set at processing block 508 at processing block 516 and determines whether the data should be written to the device at decision box 518. judge. This determination is written to the device for security (eg: writing some data to the device but not allowing writing other data) or correctness (eg: if required by virtualization) May need to be changed before).

  If the determination made in decision box 518 is positive, processing logic in the VMM writes data to the device at processing block 520. Alternatively, the VMM may modify the data before writing it to the device, for example to ensure that certain properties regarding security, performance, or device virtualization are appropriate.

  Thus, a method and apparatus have been described that provides support for virtual machine single-step functionality in a virtual machine environment. The above description is intended to illustrate the present invention and is not intended to limit the present invention. Many other embodiments will be apparent to those of skill in the art upon reading the above description. Thus, the scope of the present invention is determined by the full scope of the appended claims and their equivalents.

Claims (21)

A step performed by single-step logic,
Receiving a request to transition control to the virtual machine (VM) from virtual machine monitor (VMM),
A step of determining that the single-step indicator is set to single-step values,
In response to determining that the single step indicator is set to a single step value,
The method comprising transferring control to the VM,
If the execution of the first instruction of the VM is successful, after the successful completion of said execution of the first instruction, the steps of transferring control to the VMM,
If the execution of the first instruction of the VM did not complete successfully due to a current fault caused by the execution of the first instruction , the current fault is related to a transfer of control to the VMM. Do the same steps for determining whether or not to,
A step of notifying that the current fault control in the VMM by the transition to and transfers control to the VMM if relevant the current fault control to the VMM is migrated to the VMM,
When the current fault is not related to the transfer of control to the VMM, and the stage to send the current of the fault to the VM, if the is the sending of the current fault was successful, the corresponding fault handler Before the instruction is executed , control is transferred to the VMM, and the value of the single step indicator indicates a single step value to notify the VMM that control has been transferred to the VMM. Stage and
And the stage of
A method comprising:   The method of claim 1, wherein the transfer of control to the VMM comprises notifying the VMM that control has been transferred to the VMM according to a current value of the single step indicator. A step of determining that said sending of the current fault is unsuccessful,
A step determining whether a new fault is associated with transfer of control to the VMM,
Wherein when a new fault is associated with transfer of control to the VMM, the VMM to transfers control, the new fault in the claims, further comprising the steps of: notifying the VMM that control the VMM is migrated The method according to 1. A step determining that the new fault is not associated with the transition of control to the VMM,
Comprising the steps of: sending the new fault to the VM,
If the delivery of the new fault is successful, control is transferred to the VMM before the corresponding fault handler instruction is executed, and control is transferred to the VMM according to the current value of the single step indicator. the method of claim 3, further comprising the steps of: notifying the VMM that was. Before transfers control to the VM, the step of determining that the request to transfer control to the VM is associated with vectorized fault is sent to the VM,
When transfers control to the VM, the steps of sending the vectorized fault to the VM,
If the sending of the vectorized fault is successful,
Corresponding to migrate vectorized controlled by the VMM to execute instructions fault handler, the method comprising: notifying that control the VMM is migrated by the current value of the single-step indicator to the VMM The method according to any one of claims 1 to 4, further comprising: A step of determining that the sending of the vectorized fault has failed,
A step determining whether a new fault is associated with transfer of control to the VMM,
If the new fault is associated with the transfer of control to the VMM, the steps of notifying that the VMM to transfers control, the control to the VMM by new fault is migrated to the VMM,
If the new fault is not related to the transfer of control to the VMM, send the new fault to the VM and transfer control to the VMM before executing the corresponding fault handler instruction; · the method of claim 5, the current value further comprises the steps of notifying the VMM that control the VMM is migrated indicator. Before transfers control to the VM, determining if said request to transition control to the VM is associated with inactive active state of the processor,
The method comprising transferring control to the VM,
The method according to any one of claims 1 to 6, further comprising the step of restraining the transfer of control to the VMM until after the occurrence of a break event.   8. If the break event is associated with a transfer of control to the VMM, further comprising transferring control to the VMM and notifying the VMM that control has been transferred to the VMM due to the break event. The method described in 1. A step of determining that the break event is not associated with transfer of control to the VMM,
Comprising the steps of: sending the break event in the VM,
If the sending of the break event ends normally, control is transferred to the VMM before executing the instruction of the corresponding handler, and control is transferred to the VMM according to the current value of the single step indicator. comprising the steps of: notifying the VMM,
If the sending of the break event fails, determining if the new fault is associated with transfer of control to the VMM,
If the new fault is associated with the transfer of control to the VMM, the steps of notifying that the VMM to transfers control, the control to the VMM by new fault is migrated to the VMM,
If the new fault is not related to the transfer of control to the VMM, control is transferred to the VMM before executing the corresponding fault handler instruction, and the current value of the single-step indicator returns to the VMM. the method according to claim 7 or 8 comprising that the control has been migrated further comprising: notifying the VMM. The single-step indicators A method according to any one of claims 1 to 9 included in the virtual machine control structure (VMCS). Accessing the save operation status indicator before transferring control to the VMM, and further comprising saving an indicator of the operation status of the VM if the save operation status indicator is set to a save value. The method according to any one of 1 to 10. And detecting an event having a higher priority than the single step after the execution of the first instruction,
Said high priority events and determining whether to be processed by the VMM,
If the high priority event is handled by the VMM, the steps of notifying that the VMM to transfers control, control on the VMM by the high priority event is migrated to the VMM,
If the high priority event is not handled by the VMM, further comprising the steps of transferring control of the standby VM exit indicator set to a single step value, the software designated to handle the higher priority event 12. A method according to any one of claims 1 to 11.   The method of claim 12, wherein the designated software sends a standby VM exit to the VMM with a current value of the single step indicator.   The method according to claim 12 or 13, wherein the designated software requests a processor to send a standby VM exit to the VMM according to a current value of the standby VM exit indicator. And detecting that after the specified software has completed the operation, the current value of the single-step indicator VM exit is waiting,
The method comprising transfer control to the VMM,
The method according to any one of the single-step indicator that control the VMM is migrated by the current value of claims 12, further comprising the steps of: notifying the VMM 14. A virtual machine monitor (VMM),
A data structure controlled by the VMM that stores a single step indicator;
A request is received from the VMM to transfer control to a virtual machine (VM), it is determined that the single step indicator is set to a single step value, and the single step indicator is set to a single step value. If the execution of the first instruction of the VM ends normally in response to determining that the first instruction of the VM is transferred, control is transferred to the VMM after the normal end of the first instruction. If the execution was not successful due to a current fault caused by the execution of the first instruction, determine whether the current fault is related to a transfer of control to the VMM;
If the current fault is related to the transfer of control to the VMM, transfer control to the VMM to notify the VMM that control has been transferred to the VMM due to the current fault;
If the current fault is not related to the transfer of control to the VMM, send the current fault to the VM, and if the delivery of the current fault is successful, the corresponding fault handler instruction Control to the VMM before execution is performed, and the value of the single step indicator indicates a single step value, thereby notifying the VMM that control has been transferred Single step・ A device with logic.   The apparatus of claim 16, wherein the single step logic notifies the VMM that control has been transferred to the VMM according to a current value of the single step indicator. A memory for storing one or more indicators;
Using the one or more indicators for the purpose of determining whether a single step of a virtual machine (VM) is indicated, and determining that the one or more indicators indicate a single step. In response, if the first instruction of the VM is executed and the execution of the first instruction is normally completed, control is transferred to a virtual machine monitor (VMM) after the normal completion of the execution of the first instruction And if the execution of the first instruction of the VM was not successful due to a current fault caused by the execution of the first instruction , the current fault is related to the transfer of control to the VMM. Determine whether to
If the current fault is related to the transfer of control to the VMM, transfer control to the VMM to notify the VMM that control has been transferred to the VMM due to the current fault;
If the current fault is not related to the transfer of control to the VMM, send the current fault to the VM, and if the delivery of the current fault is successful, the corresponding fault handler instruction is Transferring control to the VMM before it is executed and notifying the VMM that control has been transferred to the VMM by the value of the one or more indicators indicating a single step value ; And a processor connected to memory. The system of claim 18, wherein the processor notifies the VMM that control has been transferred to the VMM because the value of the one or more indicators indicates a single step value . A machine readable medium recording instructions,
When the instruction group is executed by the processing system , the processor of the processing system
Receives a request from the virtual machine monitor (VMM) to transfer control to the virtual machine (VM),
In response to determining that the single step indicator is set to a single step value and determining that the single step indicator is set to a single step value,
Transfer control to the VM,
If execution of the first instruction of the VM is successful, control is transferred to the VMM after the normal completion of the execution of the first instruction, and the execution of the first instruction of the VM is If the current fault caused by the execution of the instruction did not complete successfully, determine whether the current fault is related to a transfer of control to the VMM;
If the current fault is related to the transfer of control to the VMM, transfer control to the VMM to notify the VMM that control has been transferred to the VMM due to the current fault; If a fault is not associated with the transfer of control to the VMM, send the current fault to the VM, and if the delivery of the current fault is successful, the corresponding fault handler instruction is executed. A medium that has previously transferred control to the VMM and notifies the VMM that control has been transferred because the value of the single step indicator indicates a single step value . The control transfer to the VMM comprises notifying the VMM that control has been transferred to the VMM by the value of the single step indicator indicating a single step value. The machine-readable medium described.

Images